Fix #120, fix #125, fix #134 - Add .griveignore

README.md

@@ -85,6 +85,7 @@ Grive uses cmake to build. Basic install sequence is
 
 ### Grive2 v0.5.1-dev
 
+- support for .griveignore
 - no-remote-new and upload-only modes
 - ignore regexp does not persist anymore (note that Grive will still track it to not
   accidentally delete remote files when changing ignore regexp)
@@ -137,3 +138,21 @@ New features:
 - #87: support for revisions
 - #86: ~~partial sync (contributed by justin at tierramedia.com)~~ that's not partial sync,
   that's only support for specifying local path on command line
+
+## .griveignore
+
+Rules are similar to Git's .gitignore, but may differ slightly due to the different
+implementation.
+
+- lines that start with # are comments
+- leading and trailing spaces ignored unless escaped with \
+- non-empty lines without ! in front are treated as "exclude" patterns
+- non-empty lines with ! in front are treated as "include" patterns
+  and have a priority over all "exclude" ones
+- patterns are matched against the filenames relative to the grive root
+- a/**/b matches any number of subpaths between a and b, including 0
+- **/a matches `a` inside any directory
+- b/** matches everything inside `b`, but not b itself
+- \* matches any number of any characters except /
+- ? matches any character except /
+- .griveignore itself isn't ignored by default, but you can include it in itself to ignore

grive/doc/grive.1

@@ -46,10 +46,6 @@ to download only files that are changed in Google Drive and already exist locall
 \fB\-h\fR, \fB\-\-help\fR
 Produces help message
 .TP
-\fB\-\-ignore\fR <perl_regexp>
-Ignore files with relative paths matching this Perl Regular Expression.
-Value is remembered for next runs.
-.TP
 \fB\-l\fR <filename>, \fB\-\-log\fR <filename>
 Write log output to
 .I <filename>
@@ -81,6 +77,37 @@ Print ASCII progress bar for each downloaded/uploaded file.
 \fB\-V\fR, \fB\-\-verbose\fR
 Verbose mode. Enables more messages than usual.
 
+.SH .griveignore
+.PP
+You may create .griveignore in your Grive root and use it to setup
+exclusion/inclusion rules.
+.PP
+Rules are similar to Git's .gitignore, but may differ slightly due to the different
+implementation.
+.IP \[bu]
+lines that start with # are comments
+.IP \[bu]
+leading and trailing spaces ignored unless escaped with \\
+.IP \[bu]
+non-empty lines without ! in front are treated as "exclude" patterns
+.IP \[bu]
+non-empty lines with ! in front are treated as "include" patterns
+and have a priority over all "exclude" ones
+.IP \[bu]
+patterns are matched against the filenames relative to the grive root
+.IP \[bu]
+a/**/b matches any number of subpaths between a and b, including 0
+.IP \[bu]
+**/a matches `a` inside any directory
+.IP \[bu]
+b/** matches everything inside `b`, but not b itself
+.IP \[bu]
+* matches any number of any characters except /
+.IP \[bu]
+? matches any character except /
+.IP \[bu]
+\[char46]griveignore itself isn't ignored by default, but you can include it in itself to ignore
+
 .SH AUTHORS
 .PP
 Current maintainer is Vitaliy Filippov.

grive/src/main.cc

@@ -124,15 +124,22 @@ int Main( int argc, char **argv )
 		( "no-remote-new,n", "Download only files that are changed in Google Drive and already exist locally" )
 		( "dry-run",	"Only detect which files need to be uploaded/downloaded, "
 						"without actually performing them." )
-		( "ignore",		po::value<std::string>(), "Perl RegExp to ignore files (matched against relative paths)." )
 		( "upload-speed,U", po::value<unsigned>(), "Limit upload speed in kbytes per second" )
 		( "download-speed,D", po::value<unsigned>(), "Limit download speed in kbytes per second" )
 		( "progress-bar,P", "Enable progress bar for upload/download of files")
 	;
 	
 	po::variables_map vm;
-	po::store(po::parse_command_line( argc, argv, desc), vm );
+	try
-	po::notify(vm);
+	{
+		po::store( po::parse_command_line( argc, argv, desc ), vm );
+	}
+	catch( po::error &e )
+	{
+		std::cerr << "Options are incorrect. Use -h for help\n";
+		return -1;
+	}
+	po::notify( vm );
 	
 	// simple commands that doesn't require log or config
 	if ( vm.count("help") )
@@ -148,9 +155,9 @@ int Main( int argc, char **argv )
 	}
 
 	// initialize logging
-	InitLog(vm) ;
+	InitLog( vm ) ;
 	
-	Config config(vm) ;
+	Config config( vm ) ;
 	
 	Log( "config file name %1%", config.Filename(), log::verbose );
 

libgrive/src/base/Drive.cc

@@ -41,15 +41,10 @@
 
 namespace gr {
 
-namespace
-{
-	const std::string state_file = ".grive_state" ;
-}
-
 Drive::Drive( Syncer *syncer, const Val& options ) :
 	m_syncer	( syncer ),
 	m_root		( options["path"].Str() ),
-	m_state		( m_root / state_file, options ),
+	m_state		( m_root, options ),
 	m_options	( options )
 {
 	assert( m_syncer ) ;
@@ -72,7 +67,7 @@ void Drive::FromChange( const Entry& entry )
 
 void Drive::SaveState()
 {
-	m_state.Write( m_root / state_file ) ;
+	m_state.Write() ;
 }
 
 void Drive::DetectChanges()

libgrive/src/base/State.cc

@@ -32,17 +32,28 @@
 
 namespace gr {
 
-State::State( const fs::path& filename, const Val& options  ) :
+const std::string state_file = ".grive_state" ;
+const std::string ignore_file = ".griveignore" ;
+const int MAX_IGN = 65536 ;
+const char* regex_escape_chars = ".^$|()[]{}*+?\\";
+const boost::regex regex_escape_re( "[.^$|()\\[\\]{}*+?\\\\]" );
+
+inline std::string regex_escape( std::string s )
+{
+	return regex_replace( s, regex_escape_re, "\\\\&", boost::format_sed );
+}
+
+State::State( const fs::path& root, const Val& options  ) :
+	m_root		( root ),
 	m_res		( options["path"].Str() ),
 	m_cstamp	( -1 )
 {
-	Read( filename ) ;
+	Read() ;
-	
+
 	// the "-f" option will make grive always think remote is newer
 	m_force = options.Has( "force" ) ? options["force"].Bool() : false ;
-	
+
 	std::string m_orig_ign = m_ign;
-	m_ign = "";
 	if ( options.Has( "ignore" ) && options["ignore"].Str() != m_ign )
 		m_ign = options["ignore"].Str();
 	else if ( options.Has( "dir" ) )
@@ -52,8 +63,7 @@ State::State( const fs::path& filename, const Val& options  ) :
 		if ( !m_dir.empty() )
 		{
 			// "-s" is internally converted to an ignore regexp
-			const boost::regex esc( "[.^$|()\\[\\]{}*+?\\\\]" );
+			m_dir = regex_escape( m_dir );
-			m_dir = regex_replace( m_dir, esc, "\\\\&", boost::format_sed );
 			size_t pos = 0;
 			while ( ( pos = m_dir.find( '/', pos ) ) != std::string::npos )
 			{
@@ -66,7 +76,7 @@ State::State( const fs::path& filename, const Val& options  ) :
 	}
 
 	m_ign_changed = m_orig_ign != "" && m_orig_ign != m_ign;
-	m_ign_re = boost::regex( m_ign.empty() ? "^\\.(grive|grive_state|trash)" : ( m_ign+"|^\\.(grive|grive_state|trash)" ) );
+	m_ign_re = boost::regex( m_ign.empty() ? "^\\.(grive$|grive_state$|trash)" : ( m_ign+"|^\\.(grive|grive_state|trash)" ) );
 }
 
 State::~State()
@@ -277,27 +287,92 @@ State::iterator State::end()
 	return m_res.end() ;
 }
 
-void State::Read( const fs::path& filename )
+void State::Read()
 {
 	try
 	{
-		File file( filename ) ;
+		File st_file( m_root / state_file ) ;
-
+		m_st = ParseJson( st_file );
-		m_st = ParseJson( file );
-		m_ign = m_st.Has( "ignore_regexp" ) ? m_st["ignore_regexp"].Str() : std::string();
-
 		m_cstamp = m_st["change_stamp"].Int() ;
 	}
 	catch ( Exception& )
 	{
 	}
+
+	try
+	{
+		File ign_file( m_root / ignore_file ) ;
+		char ign[MAX_IGN] = { 0 };
+		int s = ign_file.Read( ign, MAX_IGN-1 ) ;
+		ParseIgnoreFile( ign, s );
+	}
+	catch ( Exception& e )
+	{
+	throw e;
+	}
 }
 
-void State::Write( const fs::path& filename )
+bool State::ParseIgnoreFile( const char* buffer, int size )
+{
+	const boost::regex re1( "/\\\\\\*\\\\\\*$" );
+	const boost::regex re2( "^\\\\\\*\\\\\\*/" );
+	const boost::regex re3( "([^\\\\](\\\\\\\\)*)/\\\\\\*\\\\\\*/" );
+	const boost::regex re4( "([^\\\\](\\\\\\\\)*|^)\\\\\\*" );
+	const boost::regex re5( "([^\\\\](\\\\\\\\)*|^)\\\\\\?" );
+	std::string exclude_re, include_re;
+	int prev = 0;
+	for ( int i = 0; i <= size; i++ )
+	{
+		if ( buffer[i] == '\n' || ( i == size && i > prev ) )
+		{
+			while ( prev < i && ( buffer[prev] == ' ' || buffer[prev] == '\t' || buffer[prev] == '\r' ) )
+				prev++;
+			if ( buffer[prev] != '#' )
+			{
+				int j;
+				for ( j = i-1; j > prev; j-- )
+					if ( buffer[j-1] == '\\' || ( buffer[j] != ' ' && buffer[j] != '\t' && buffer[j] != '\r' ) )
+						break;
+				std::string str( buffer+prev, j+1-prev );
+				bool inc = str[0] == '!';
+				if ( inc )
+					str = str.substr( 1 );
+				str = regex_escape( str );
+				str = regex_replace( str, re1, "/.*", boost::format_perl );
+				str = regex_replace( str, re2, ".*/", boost::format_perl );
+				str = regex_replace( str, re3, "$1/(.*/)*", boost::format_perl );
+				str = regex_replace( str, re4, "$1[^/]*", boost::format_perl );
+				std::string str1;
+				while (1)
+				{
+					str1 = regex_replace( str, re5, "$1[^/]", boost::format_perl );
+					if ( str1.size() == str.size() )
+						break;
+					str = str1;
+				}
+				if ( !inc )
+					exclude_re = exclude_re + ( exclude_re.size() > 0 ? "|" : "" ) + str;
+				else
+					include_re = include_re + ( include_re.size() > 0 ? "|" : "" ) + str;
+			}
+			prev = i+1;
+		}
+	}
+
+	if ( exclude_re.size() > 0 )
+	{
+		m_ign = "^" + ( include_re.size() > 0 ? "(?!" + include_re + ")" : std::string() ) + "(" + exclude_re + ")$";
+		return true;
+	}
+	return false;
+}
+
+void State::Write()
 {
 	m_st.Set( "change_stamp", Val( m_cstamp ) ) ;
 	m_st.Set( "ignore_regexp", Val( m_ign ) ) ;
 	
+	fs::path filename = m_root / state_file ;
 	std::ofstream fs( filename.string().c_str() ) ;
 	fs << m_st ;
 }

libgrive/src/base/State.hh

@@ -42,15 +42,15 @@ public :
 	typedef ResourceTree::iterator iterator ;
 
 public :
-	explicit State( const fs::path& filename, const Val& options ) ;
+	explicit State( const fs::path& root, const Val& options ) ;
 	~State() ;
 	
 	void FromLocal( const fs::path& p ) ;
 	void FromRemote( const Entry& e ) ;
 	void ResolveEntry() ;
 	
-	void Read( const fs::path& filename ) ;
+	void Read() ;
-	void Write( const fs::path& filename ) ;
+	void Write() ;
 
 	Resource* FindByHref( const std::string& href ) ;
 	Resource* FindByID( const std::string& id ) ;
@@ -64,6 +64,7 @@ public :
 	void ChangeStamp( long cstamp ) ;
 
 private :
+	bool ParseIgnoreFile( const char* buffer, int size ) ;
 	void FromLocal( const fs::path& p, Resource *folder, Val& tree ) ;
 	void FromChange( const Entry& e ) ;
 	bool Update( const Entry& e ) ;
@@ -72,6 +73,7 @@ private :
 	bool IsIgnore( const std::string& filename ) ;
 	
 private :
+	fs::path			m_root ;
 	ResourceTree		m_res ;
 	int					m_cstamp ;
 	std::string			m_ign ;